Gastroenteropancreatic neuroendocrine tumors: impact of consistent contrast agent selection on radiologists' confidence in hepatic lesion assessment on restaging MRIs

Purpose: To evaluate the impact of contrast agent selection on radiologists' confidence in assessing liver lesions on follow-up magnetic resonance imaging (MRI) studies in patients with neuroendocrine tumors. Methods: This Institutional Review Board-approved, retrospective study performed at a tertiary cancer center and a quaternary care urban academic hospital included all 694 follow-up abdominal MRI studies from 179 patients with gastroenteropancreatic neuroendocrine tumor performed from 01/01/2010 to 05/31/2015. Primary outcome measure was radiologists' confidence in assessing liver lesions on follow-up MRI. MRI reports were reviewed to abstract radiologists' confidence, classified as equivocal if any equivocal connotation (mention of limitation due to differences in contrast agent or follow-up recommendation with specific contrast agent) was present; or unequivocal if a precise, confident comparison to prior was documented without the use of ambiguous terms. A fellowship-trained radiologist separately evaluated 100 randomly selected reports and images to calculate interobserver agreement with the report classification (equivocal vs. unequivocal) and with the original MRI report, respectively. Chi-square test was used to compare the proportion of equivocal reports when same or different contrast agent was used for successive examinations. Results: Rates of equivocal reports were higher when different contrast agents were used for successive examinations compared to examinations with same contrast agent (13.2% [21/159] vs. 1.8% [10/535]; p < 0.0001). There was very good interobserver agreement for assessment of radiologist confidence (kappa = 0.92 for report review, kappa = 0.82 for image review). Conclusions: Consistent use of contrast agent for followup MRIs allows more confident assessment of liver lesions in patients with neuroendocrine tumors.